Effects of dietary levels of 20:5n 3 and 22:6n 3 on tissue lipid composition in juvenile Atlantic salmon, Salmo salar, with emphasis on brain and eye

Transcription

1 Aquaculture Nutrition ; Effects of dietary levels of 20:5n 3 and 22:6n 3 on tissue lipid composition in juvenile Atlantic salmon, Salmo salar, with emphasis on brain and eye T. BRODTKORB Institute of Nutrition, Directorate of Fisheries, Bergen, Norway G. ROSENLUND Nutreco Aquaculture Research Centre A/S, Forus, Norway Ø. LIE Institute of Nutrition, Directorate of Fisheries, Bergen, Norway Abstract Four dietary groups of juvenile Atlantic salmon, Salmo salar L., each with three replicates, were fed diets with increasing levels of docosahexaenoic acid (22:6n 3; DHA) and eicosapentaenoic acid (20:5n 3; EPA). Fatty acid composition of brain and eye was determined at the start and approximately every 3 weeks during the experimental period, and fatty acid composition of liver and fillet was determined in fish from the final sampling. Lipid class composition of brain and eye, and fatty acid composition of these lipid classes was determined at the end of the experiment. There was no effect of increasing dietary DHA content on fatty acid composition, lipid class composition or DHA levels in the lipid classes in the juvenile Atlantic salmon brain. The increasing dietary EPA content, however, was reflected in both the total fatty acid composition and in the EPA content in neutral lipids, phosphatidylcholine (PC), phosphatidylethanolamine (PE) and phosphatidylinositol (PI). A minor effect of the increasing dietary DHA content was found in the lipid composition of the juvenile salmon eye. Both EPA and 18:2n 6 levels in eye, however, clearly reflected the increasing and decreasing, respectively, dietary levels of these two fatty acids. The dietary EPA levels also affected the EPA levels in neutral lipids, PC, PE, PI and PS (phosphatidylserine) in the juvenile salmon eye. The results demonstrate that these dietary levels of DHA had no effect on brain lipid composition and only a minor effect on eye lipid composition. Furthermore, the dietary EPA levels significantly affected the lipid composition of both brain and eye. The fillet fatty acid composition reflected the dietary fatty acid composition, except for the DHA/EPA ratio, which was reversed in fillet compared with that in the diets. The liver fatty acid composition was also affected by the increasing dietary EPA and DHA levels. KEY WORDS: brain, dietary influence, eye, fatty acid composition, fillet, juvenile Atlantic salmon, lipid class composition, liver, 22:6n 3 (DHA), 20:5n 3 (EPA) Received 24 June 1996, accepted 23 October 1996 Correspondence: T. Brodtkorb, Institute of Nutrition, Directorate of Fisheries, PO Box 185, N-5002 Bergen, Norway Introduction The two (n 3) fatty acids eicosapentaenoic acid, 20:5n 3 (EPA) and docosahexaenoic acid, 22:6n 3 (DHA), are abundant in both marine and freshwater food webs (Sargent et al. 1989), and are only to a small extent synthesized from 18:3n 3 in fish (Sargent et al. 1989). It is essential to provide these fatty acids through the diet in order to achieve optimal growth and development (Sargent et al. 1989). The retina and brain of higher vertebrates are, relative to other higher vertebrate tissues, characteristically rich in 22:6n 3 (Tinoco 1982). This is also the case in fish (Tocher & Harvie 1988). Good vision is of particular relevance in larval fish for identifying, hunting and capturing live prey. Reduced twilight vision was demonstrated in DHA-deficient juvenile herring, Clupea harengus L. (Bell et al. 1995). It was recently reported that the deficit of di-22:6n 3 in retina in 22:6n 3 deficient fish is compensated for by increased levels of 20:5n 3 and 22:5n 3, especially in PE (phosphatidylethanolamine) and PS (phosphatidylserine) (Bell et al. 1995). Depletion of 22:6n 3 in eye and brain results in reduced visual and cognitive abilities in mammals (Crawford 1990). This indicates a specialized role of 22:6n 3 for normal development and function of eye and brain in mammals and fish. DHA is present in particularly high concentrations in the synaptic membranes of the brain (Connor et al. 1992) with the highest concentration of 22:6n 3 found in the grey matter of the cerebral cortex (Tinoco 1982). The high concentration of long-chain fatty 1997 Blackwell Science Ltd Effects on Atlantic Salmon fed what starch 175

2 176 T. Brodtkorb et al. acids (LCFA), especially of DHA, in central nervous system (CNS) membranes may indicate that the membrane LCFA composition is important for of the normal functioning of excitable tissue (Innis 1991). The biochemical significance of this enrichment, however, is not well understood (Salem et al. 1986). Mammalian retina and brain probably have mechanisms for selective uptake of C22 PUFA (polyunsaturated fatty acids) (Wang et al. 1992). However, cultured brain cells from turbot, Scophthalmus maximus (L.), showed no preferential uptake of 22:6n 3 into major phospholipids, when compared with 18:3n 3 (Tocher et al. 1992). The authors concluded that the specificity of the brain fatty acyl transferases incorporating PUFA into major phosphoacylglycerols are very low, and thereby the brains of these marine carnivore species are highly vulnerable to changes in circulating PUFA depending upon changes in dietary PUFA. Developing turbot incorporated DHA into brain in preference to other fatty acids (Mourente & Tocher 1992). This was also demonstrated in rats (Anderson & Connor 1988). Furthermore, 22:6n 3 was strongly retained in brain lipids of sea bass, Dicentrarchus labrax (L.) (Pagliarani et al. 1986) and in brain and retinas of cod, Gadus morhua L. (Tocher & Harvie 1988) and rainbow trout, Oncorhynchus mykiss (Walbaum) (Bell & Tocher 1989). Di-22:6n 3 phospholipid species are more abundant in fish than in mammals (Bell & Dick 1991). The aim of the present study was to examine the relationship between fillet, liver, brain and eye lipids in juvenile Atlantic salmon, Salmo salar L., with respect to the incorporation of dietary fatty acids, and especially of EPA and DHA, when fed varying dietary levels of these fatty acids. Materials and methods Fish and diets The experiment was carried out from January to April 1995 at Nutreco ARC s research station, Stavanger, Norway. Juvenile Atlantic salmon, ready for exogenous feeding (mean weight: g) were distributed into m tanks (2000 in each). The four experimental diets (Table 1) with increasing levels of Table 1 Composition (g kg 1 ) of the experimental diets Feed compositon Diet 1 Diet 2 Diet 3 Diet 4 Fish meal Soya concentrate Soya oil Capelin oil DHA-EPA-acetate Vitamin and mineral mix Premix Carbohydrate source etc. EPA and DHA were fed in excess to triplicate tanks. Graded levels of DHA and EPA (Table 2) were achieved by replacing part of the capelin and soya oil by a DHA-EPA-acetate (T.FP.01.G, Pronova, 85% purity). The diets were produced by T. Skretting A/S, Stavanger, Norway. The diets were formulated to contain 520 g kg 1 protein, 200 g kg 1 fat and 140 g kg 1 NFE (nitrogen free extract). Mortalities were recorded and removed daily. Weight was determined by bulk weighing of all the fish at the beginning of the trial and thereafter approximately every 3 weeks troughout the trial. The mean temperature and O 2 -level during the experimental period were 9.8 C ± 0.6 and 11 mg L 1 ± 1.0 (mean ± SD), respectively. The tanks were exposed continuously to light. Sampling procedure Samples were taken from each batch of all feeds and stored at 20 C. Fish were sampled at the start and thereafter every 3 weeks. Randomly selected fish were anaesthetized and killed by adding a saturated solution of metacainum. The sampled fish were bulk weighed and immediately frozen on dry ice and stored at 80 C. Pooled samples of brain and eye from five fish were collected from the first four samplings, whereas pooled samples of brain, eye, liver and fillet (without skin) from 10 fish were collected from the final sampling. Analytical procedures Pooled samples of brain, eye, liver, fillet and feed were homogenized, extracted with chloroform/methanol (2:1, v/v) and 19:0 was added as internal standard. Then the samples were filtrated, saponified and esterified in 12% BF 3 in methanol. Fatty acid composition of total lipids were analysed using methods described by Lie & Lambertsen (1991) where the methyl esters were separated using a Carlo Erba gas chromatograph ( cold on column injection, C/min C/min C/min 220 C), equipped with a 50 m CP-sil 88 (Chromopack) fused silica capillary column (i.d mm). The fatty acid composition (weight percentage) was calculated using an integrator (Turbochrom Navigator, Version 4.0), connected to the GLC and identification ascertained by standard mixtures of methyl esters (Nu-Chek, Elyian, MN, USA). The lipids from the brain and eye samples from the final sampling were extracted and the neutral lipids, phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylinositol (PI) and phosphatidylserine (PS) were separated by HPLC according to Lie & Lambertsen (1991) using a Constametric II solvent-delivery system and at 205 nm with a variablewavelength spectrophotometer (LDC Spectromonitor III). The

3 Fatty acid effects on salmon brain and eye 177 Table 2 Fatty acid composition, wt% and as mg g 1 diet, of the experimental diets. Data are presented as mean (n = 2) Diet 1 Diet 2 Diet 3 Diet 4 % mg g 1 % mg g 1 % mg g 1 % mg g 1 14: : : : saturated :1n :1n :1n :1n :1n :1n :1n monoenes :2n :4n n :3n :4n :5n :5n :6n n n 3/n column (25 cm 0.46 cm i.d.) was packed with silica gel (LiChrosorb 5 µm, Merck). The lipid fractions were evaporated to dryness, saponified, esterified in 12% BF 3 in methanol and the fatty acid compositions were determined according to Lie & Lambertsen (1991). The lipid classes were separated and quantified according to the method described by Rønnestad et al. (1995) using an Iatroscan thin-layer chromatography-flame ionization detector (FID) system (Iatroscan MK-5, Iatron Laboratories Inc., Tokyo) modified after Tocher et al. (1985). Statistics Differences for certain individual fatty acids (18:2n 6, 20:4n 6, 20:5n 3, 22:5n 3 and 22:6n 3) between the groups in the final sampling were examined by using a Kruskal Wallis ANOVA by ranks and intergroup differences using a non-parametric Mann Whitney U-test (CSS:Statistica 4.5, Statsoft, Inc., USA 1993), and differences between groups one and four for the same five individual fatty acids within each lipid class were examined using a t-test for independent samples (CSS:Statistica 4.5, Statsoft, Inc., USA 1993). The relative data of fatty acid composition in brain, eye, liver, fillet and in the different phospholipids in brain and eye were analysed using the SIRIUS for Windows program (Version 1.2). Principal component analyses (PCA) (Wold et al. 1987) were performed in each data matrix from the respective organs. The purpose of PCA is to express the main information in the variables by a lower number of variables, the so-called principal components (PC1, PC2, ). Two significant principal components were found in each PCA. PC1 represents a vector along the longest axis in the multidimensional cloud of data points and usually explains the dominant part of the actual variance. The second component (PC2) runs perpendicular to PC1. A high positive or negative loading reveals a significant variable in the actual PCA model. Score plots from the PCA explored the main trends in the data, and their respective loadings revealed the significant fatty acids. Results Groups 1 to 4 were fed diets with increasing levels of 22:6n 3 and 20:5n 3 and decreasing levels of 18:2n 6 (Table 2). The juvenile Atlantic salmon increased their weight nearly 12-fold, from g to 1.76 g ± 0.05, during the experimental period of 93 days, and there were no significant differences in growth between the four groups. The mortality during the experimental period was 5% in all four groups.

4 178 T. Brodtkorb et al. Table 3 Fatty acid composition of eye and brain from the juvenile salmon fed increasing levels of DHA and EPA (groups 1 to 4) from the initial and final sampling. 1 Data are in wt% of total fatty acids and are shown as mean ± SEM;, < 0.1 Brain Eye Initial Final Initial Final sampling sampling (n = 3) sampling sampling (n = 3) (n = 1) (n = 1) Diet 1 Diet 2 Diet 3 Diet 4 Diet 1 Diet 2 Diet 3 Diet 4 14: ± ± ± 0.4 ± ± ± ± ± : ± ± ± ± ± ± ± ± : ± ± ± ± ± 0.4 ± ± 0.2 ± 18: ± ± ± ± ± ± ± ± 1.5 saturated ± ± ± ± ± ± ± ± :1n ± ± ± ± ± ± ± 1.8 ± :1n ± 0.7 ± ± ± ± 0.3 ± 0.3 ± 0.4 ± :1n ± 3.1 ± ± ± ± ± 2.0 ± 2.3 ± :1n ± ± ± ± ± ± ± ± :1n ± 1.5 ± ± ± ± ± ± 1.4 ± :1n ± 0.7 ± ± ± ± ± 2.0 ± 1.3 ± :1n ± ± ± ± ± ± ± 0.8 ± 0.1 monoenes ± ± ± ± ± ± ± ± :2n ± ± ± ± ± 1.9 ade 16.0 ± 0.4 d 13.2 ± 0.2 e 7.9 ± 2.5 b 20:4n ± 1.3 ± 1.2 ± 1.2 ± ± ± 1.2 ± 1.4 ± 0.1 n ± ± ± ± ± ± ± ± :3n ± 0.4 ± ± ± ± ± ± 1.2 ± :4n ± 0.3 ± 0.4 ± 0.3 ± ± 0.7 ± 0.6 ± 0.6 ± :5n ± 0.2 a 6.7 ± 0.1 a 7.2 ± 0.3 a 7.9 ± 0.1 b ± 1.2 acd 5.9 ± c 8.6 ± 0.1 b 10.8 ± 2.0 bd 22:5n ± 0.1 a 2.5 ± 0.2 ad 3.0 ± 0.1 b 3.3 ± c ± 0.3 acd 2.1 ± c 2.8 ± b 3.3 ± 0.3 bd 22:6n ± ± ± ± ± ± ± ± 1.7 n ± ± ± ± ± ± ± ± 1.7 n 3/n ± ± ± ± ± ± ± ± 1.9 rest FA Small letters indicate significant differences (P 0.05) among groups for individual fatty acids in the final sampling. Identical letter, not significantly different. Where no letters are indicated, no significant differences were observed (only 18:2n 6, 20:4n 6, 20:5n 3, 22:5n 3 and 22:6n 3 are tested for significant differences). Fatty acid composition The fatty acid composition was analysed in brain and eye lipids in fish from all samplings, and the fatty acid compositions from the initial and final sampling are presented in Table 3. The fatty acid compositions in brain and eye lipids were similar before feeding. At the initial sampling, 22:6n 3 was the most abundant fatty acid in both organs, with 25.9% in the brain and 24.5% in the eye. The relative 22:6n 3 content in brain increased during the first 3 weeks ( 7% increase) in all groups Fig. 1A). The relative level of both 20:5n 3 (Fig. 1B) and 22:5n 3 were significantly different between the groups at the end of the experiment (Table 3). Furthermore, the relative 18:2n 6 content in brain was highest in group 1 and lowest in group 4, but the levels were not significantly different and stayed quite stable within each group throughout the experiment. The relative 22:6n 3 level in eye lipids increased by % during the first period (Fig. 1C). This was followed by a slow decrease, especially for groups 1 and 2, during the rest of the experiment. The 20:4n 6 and 22:5n 3 content in eye both decreased, while the 18:2n 6 content increased markedly in all groups during the experimental period. The 20:5n 3 content in eye lipids (Fig. 1D) was doubled in group 4 compared with group 1, which was the same ratio as in their respective diets. Furthermore, the relative levels of 18:2n 6, 20:5n 3 and 22:5n 3 between the groups were significantly different at the end of the experiment. The relative fatty acid composition of liver and fillet in fish from the final sampling is shown in Table 4. The relative level of 22:6n 3 increased by 7.6% and 8.6% in liver and fillet, respectively, from group 1 to group 4. Furthermore, the relative level of

5 Fatty acid effects on salmon brain and eye 179 Figure 2 Lipid class composition (mg g 1 ) of brain and eye from the final sampling. Groups 1 4 were fed diets with increasing EPA and DHA levels. Data are shown as mean ± SEM (n = 3); vertical scales differ. Figure 1 Relative level (wt%) of 22:6n 3 (A) and 20:5n 3 (B) in brain and of 22:6n 3 (C) and 20:5n 3 (D) in eye from fish fed increasing levels of DHA and EPA (1 4) throughout the experiment. Data are shown as mean ± SEM (n = 3); vertical scales differ. 20:5n 3 and 22:5n 3 increased by 7% and 1.7%, respectively, in liver and by 7.9% and 2.5%, respectively, in fillet from group 1 to group 4. The 18:2n 6 content decreased in both organs from group 1 to group 4. Lipid classes The lipid classes are presented graphically in Fig. 2. The different 22:6n 3 and 20:5n 3 levels in the experimental diets had no significant effect on the brain and eye lipid class composition. Triacylglycerol (TAG) was the most abundant lipid class in eye lipids, independent of diet, whilst none of the other lipid classes in the eye exceeded 5 mg g 1 wet weight tissue. In contrast, phosphatidylethanolamine (PE) was the main lipid class in the brain. The distribution of the other lipid classes in brain was PC > Chol > TAG > SM > SE > DG + PS + LPC in gradually decreasing amounts (Fig. 2). Fatty acid composition of the lipid classes Fatty acid compositions of the lipid classes were determined only in group 1 and group 4 (Tables 5, 6). The neutral lipid class in the eye was characterized by having the largest increases in 22:6n 3, 22:5n 3, 20:5n 3, 20:4n 6 and decrease in 18:2n 6 content from group 1 to group 4 compared with the phosphoacylglycerols. The phosphoacylglycerols in both organs and the neutral lipid class in brain showed little response towards the different 22:6n 3, 20:5n 3 and 18:2n 6 contents in the two diets. However, the statistical tests revealed that the 20:5n 3 level was significantly

6 180 T. Brodtkorb et al. Table 4 Final sampling. Fatty acid composition (n = 3) of liver and fillet from the juvenile salmon fed increasing levels of DHA and EPA (groups 1 to 4). All data are in wt% of total fatty acids and are shown as mean ± SEM; < 0.1 Liver Fillet Diet 1 Diet 2 Diet 3 Diet 4 Diet 1 Diet 2 Diet 3 Diet 4 14:0 2.5 ± ± ± ± 3.2 ± ± ± 2.4 ± : ± ± ± ± ± ± ± ± :0 0.3 ± 0.3 ± 0.3 ± 0.3 ± 0.2 ± 0.2 ± 0.2 ± 0.2 ± :0 5.0 ± ± ± ± 3.9 ± ± ± 3.0 ± saturated 23.3 ± ± ± ± ± ± ± ± :1n ± ± ± ± ± ± ± ± :1n ± 1.2 ± ± 0.3 ± 0.2 ± 0.2 ± 0.2 ± 0.1 ± :1n ± ± ± ± 2.2 ± 2.1 ± 1.8 ± 1.5 ± 18:1n ± ± ± ± ± ± ± ± :1n ± ± ± ± 2.0 ± ± ± 1.8 ± 22:1n ± ± ± ± ± ± ± 2.1 ± :1n ± 0.2 ± ± 0.6 ± 0.4 ± ± ± ± 0.1 monoenes 20.7 ± ± ± ± ± ± ± ± :2n ± ± ± ± ± ± ± ± :4n ± ± ± ± ± 0.8 ± 0.9 ± 1.1 ± n ± ± ± ± ± ± ± ± :3n ± 1.6 ± ± ± 2.4 ± 2.4 ± 2.2 ± 1.2 ± :4n ± 0.6 ± 0.6 ± 0.6 ± 0.7 ± 0.8 ± 0.8 ± 0.7 ± 20:5n ± ± ± ± ± ± ± ± 22:5n ± 1.2 ± ± ± ± ± 3.1 ± 3.9 ± 22:6n ± ± ± ± ± ± ± ± 0.6 n ± ± ± ± ± ± ± ± 0.5 n 3/n ± ± ± ± ± ± ± 3.6 ± 0.1 rest FA higher in group 4 than group 1 in neutral lipids, PC and PE in both brain and eye lipids, and in PI and PS in eye. Characteristic for PI in both organs was the high 20:4n 6 level (about 16.5% in eye and 9% in brain). The n 3 / n 6 ratio was generally higher in all lipid classes in brain compared with the eye lipid classes. PI had the lowest n 3 / n 6 ratio of the lipid classes in both organs due to the high 20:4n 6 content. Multivariate analysis The fatty acids (included in Tables 3 and 4, except 17:0 owing to the very low content of this fatty acid in the organs analysed; polyene fatty acids was included in the multivariate analysis) of brain, eye, liver and fillet lipids were used in multivariate analyses of the relative fatty acid data. Separate principal component analyses (PCA) were carried out on the data for the lipids of brain (Fig. 3A), eye (Fig. 3B), liver (Fig. 3C) and fillet (Fig. 3D), all including the fatty acid data of the four diets (Fe1 Fe4). The analyses of eye, brain and liver lipids revealed a high positive loading along PC1 for 22:6n 3 and n 3, and a high negative loading along PC1 for 18:2n 6 and n 6. The multivariate analysis of the fillet data, however, showed a high positive loading along PC2 for 20:5n 3 and a high negative and positive along PC2 and PC1, respectively, for 22:6n 3. The fatty acids (included in Tables 5 and 6) of brain and eye lipid classes were used in multivariate analyses of the relative fatty acid composition data. polyene fatty acids was included in the multivariate analysis. Separate PCA analyses were carried out for brain (Fig. 3E) and eye (Fig. 3F) lipid class data, each including the fatty acid data from diets 1 (Fe1) and 4 (Fe4). The analysis of the eye lipid class data showed a negative loading for 22:6n 3 and a positive loading for 16:0, 18:2n 6 and monoenes along PC2. The analysis of the brain lipid class data showed a positive loading for 22:6n 3 and n 3 along both PC1 and PC2. Discussion The four experimental groups were comparable to commercial first-feeding diets concerning growth rate and cumulative mortality, thus indicating that there were no dietary deficiencies in the experimental groups. The commercial dietary levels of EPA and DHA were in between the EPA and DHA levels of the experimental groups 2 and 3, which indicates that the juvenile salmon tolerates quite big variations in dietary EPA and DHA.

7 Fatty acid effects on salmon brain and eye 181 Table 5 Fatty acid composition (n = 3) of lipid classes from brain of the juvenile salmon fed the lowest (diet 1) and highest EPA and DHA level (diet 4). Data (wt% of total fatty acids), are shown as mean ± SEM; < 0.1 Neutral lipids PC PE PI PS Diet 1 Diet 4 (n = 2) Diet 1 Diet 4 Diet 1 Diet 4 Diet 1 Diet 4 Diet 1 (n = 2) Diet 4 14:0 0.7 ± 0.6 ± ± 0.6 ± 0.1 ± ± ± 0.3 ± 0.3 ± ± : ± ± 30.7 ± ± 12.0 ± ± ± ± ± ± :0 0.1 ± ± 0.1 ± 0.1. ± 0.3 ± 0.2 ± 0.2 ± ± 0.2 ± 0.1 ± :0 4.1 ± ± 4.1 ± ± 10.1 ± ± ± ± ± ± 1.4 saturated 35.6 ± ± 35.9 ± ± 23.0 ± ± ± ± ± ± :1n ± 1.8 ± ± 1.8 ± 0.6 ± ± 0.4 ± 0.3 ± 0.5 ± ± 16:1n ± 1.3 ± 1.3 ± 1.3 ± 0.8 ± ± 0.4 ± 0.4 ± 0.3 ± 0.3 ± 18:1n ± 1.9 ± 1.8 ± 1.9 ± 4.6 ± ± ± ± 1.5 ± ± 18:1n ± 16.0 ± ± 16.0 ± ± ± ± ± 7.7 ± ± :1n ± 0.5 ± 0.6 ± 0.5 ± 0.8 ± ± ± 0.5 ± 0.5 ± 0.4 ± 22:1n 11. ±. ±. ±. ±. ±. ±. ±. ± 0.1 ± 0.1. ± 24:1n ± 2.2 ± 2.4 ± 2.2 ± 0.1 ± ± ± ± 0.2 ± ± 0.2 monoenes 24.5 ± ± ± 23.8 ± ± ± ± ± ± ± :2n ± 0.1 a 0.7 ± b 1.3 ± 0.1 a 0.7 ± b 1.6 ± ± ± a 0.7 ± b 1.0 ± ± :2n ± 0.2 ± 0.3 ± 0.2 ± 0.5 ± ± 0.4 ± 0.2 ± 0.3 ± 0.2 ± 20:4n ± 0.6 ± 0.7 ± 0.6 ± 1.0 ± ± 9.3 ± ± ± 0.3 ± n ± ± 2.5 ± ± 3.4 ± ± ± ± ± 1.0 ± 18:3n 3. ±. ± 0.1 ± 0.1..± 0.2 ±. ±. ±. ±. ±. ± 20:4n ± 0.2 ± 0.2 ± 0.2 ± 0.3 ± 0.4 ± 0.2 ± 0.2 ± 0.2 ± 0.2 ± 20:5n ± 0.1 a 5.9 ± 0.1 b 4.9 ± 0.1 a 5.9 ± 0.1 b 5.4 ± 0.1 a 6.3 ± 0.1 b 23.2 ± ± ± 2.7 ± :5n ± ± 2.6 ± a 2.1 ± b 2.8 ± a 3.6 ± 0.1 b 1.8 ± 0.1 a 2.1 ± b 4.6 ± 0.2 a 5.8 ± 0.1 b 22:6n ± ± ± ± ± ± ± ± ± ± 2.1 n ± 36.1 ± ± ± 48.2 ± ± ± ± ± ± n 3/n ± ± ± ± ± ± ± ± 35.1 ± ± 8.0 rest FA Small letters indicate significant differences (P 0.05) between groups 1 and 4 for individual fatty acids in the final sampling. Identical letters, not significantly different. Where no letters are indicated, no significant differences were observed (only 18:2n 6, 20:4n 6, 20:5n 3, 22:5n 3 and 22:6n 3 are tested for significant differences). Fillet The score plot (PC1 vs. PC2) of the fatty acid composition in fillet and the four diets divided the fatty acid composition data into four classes (Fig. 3D) with the respective diets separate (indicated in Fig. 3D by an arrow from group 1 to group 4). This separation is due to the lower relative 20:5n 3 content, which showed a high positive loading along PC2, and higher relative 22:6n 3 content, which showed a high negative loading along PC2, in fillet compared with the respective diets. Except for these two fatty acids, the fillet fatty acid composition reflected the dietary fatty acid composition. This was expected, because fillet usually reflects the dietary fatty acid content (Lie et al. 1992; Skonberg et al. 1994). Liver The score plot (PC1 vs. PC2) of the fatty acid composition of liver and the four diets divided the material into four classes (Fig. 3C), with the diets separate from all the groups (indicated in Fig. 3C by an arrow from group 1 to group 4 and another arrow from diet 1 to diet 4). The liver samples have shifted along PC1 compared with the diets. This is due to the reversed 22:6n 3 / 20:5n 3 ratio (22:6n 3 had a high positive loading along PC1 and 20:5n 3 had high positive loading along PC2) in liver compared with the respective diets, probably caused by metabolism of these fatty acids in the liver. It is reported that 20:5n 3 is oxidized at a similar rate to palmitic acid in vitro in rats (Stray- Pedersen 1995). The current results point to an active oxidation also of 20:5n 3 in the Atlantic salmon liver. Moreover, activity of the 6- and 5-desaturation enzymes has been established in Atlantic salmon liver (Albrektsen et al. 1994) and in fibroblastlike Atlantic salmon cells (Tocher & Dick 1990a,b). The decreased 20:5n 3 and increased 22:6n 3 levels in liver compared with the diets can thus be related to the 6-desaturation, elongation and retroconversion of 20:5n 3 to 22:6n 3 in the liver. The relative contents (wt% of total fatty acids) of 18:2n 6, 20:4n 6, 20:5n 3, 22:5n 3 and 22:6n 3 in the liver of the four groups reflect the content of these fatty acids in the respective

8 182 T. Brodtkorb et al. Table 6 Fatty acid composition of the lipid classes from eye of the juvenile salmon fed the lowest EPA and DHA level (diet 1) and the highest EPA and DHA level (diet 4). Data (wt% of total fatty acids), are shown as mean ± SEM;, < 0.1 Neutral lipids PC PE PI PS Diet 1 (n = 3) Diet 4 (n = 3) Diet 1 (n = 2) Diet 4 (n = 3) Diet 1 (n = 2) Diet 4 (n = 3) Diet 1 (n = 2) Diet 4 (n = 3) Diet 1 (n = 2) Diet 4 (n = 3) 14:0 3.6 ± ± 1.3 ± ± 0.3 ± 0.2 ± 0.6 ± ± ± ± 16: ± ± ± ± ± ± ± ± ± ± :0 0.4 ± 0.3 ± ± 0.3 ± 0.3 ± 0.6 ± ± 0.6 ± ± 0.3 ± 18:0 4.3 ± ± 5.3 ± ± ± ± ± ± ± ± 0.1 saturated 22.2 ± ± ± ± ± ± ± ± ± ± :1n ± ± 1.5 ± 1.5 ± 0.6 ± 0.6 ± 0.3 ± 0.4 ± ± 1.5 ± 16:1n ± 0.2 ± 1.1 ± ± 0.5 ± 0.4 ± 0.4 ± 0.5 ± 0.6 ± ± 18:1n ± ± 1.4 ± 1.4 ± 2.8 ± ± ± 1.5 ± 1.0 ± 1.4 ± 18:1n ± ± ± ± ± ± ± 4.8 ± ± ± :1n ± 2.1 ± 0.4 ± ± 0.7 ± 0.8 ± 0.5 ± 0.5 ± 0.7 ± 0.4 ± 22:1n ± ± ± ± ± 0. ± 0.4 ± 0.1 ± ± ± :1n ± 0.9 ± 0.4 ± ± ± 0. ± 0.2 ± 0.1 ± ± 0.1 ± 0.1 monoenes 26.9 ± ± ± ± ± ± ± 7.9 ± ± ± :2n ± 0.8 a 13.8 ± b 3.0 ± 0.4 a 1.9 ± 0.1 b 3.1 ± a 2.5 ± 0.1 b 1.4 ± ± ± 0.1 a 1.9 ± :2n ± ± 0.4 ± 0.3 ± 0.6 ± 0.5 ± 0.3 ± 0.3 ± 0.4 ± 0.3 ± 20:4n ± 0.1 a 1.2 ± b 1.1 ± 1.1 ± 1.9 ± 2.3 ± ± ± ± ± n ± ± ± ± ± 5.3 ± ± ± ± ± :3n ± 2.1 ± 0.3 ± 0.3 ± 0.2 ± 0.2 ± 0.2 ± 0.1 ±. ± 0.3 ± 20:4n ± 0.8 ± 0.2 ± 0.1 ± ± 0.3 ± 0.3 ± 0.2 ± 1.6 ± ± :5n ± 0.4 a 15.0 ± 0.1 b 5.4 ± a 7.9 ± 0.1 b 4.2 ± 0.1 a 5.5 ± 0.1 b 8.7 ± 0.1 a 11.3 ± 0.2 b 1.3 ± 0.1 a 7.9 ± :5n ± 0.1 a 4.1 ± b 1.1 ± 1.4 ± ± a 2.8 ± b 1.0 ± a 1.4 ± b 2.8 ± a 1.4 ± :6n ± 0.9 a 21.9 ± 0.1 b 41.2 ± ± ± ± ± ± ± 1.2 a 42.6 ± 0.3 n ± ± ± ± ± ± ± ± ± ± 0.2 n 3/n ± ± ± ± ± ± ± ± 26.1 ± ± 0.6 rest FA Small letters indicate significant differences (P 0.05) between groups 1 and 4 for individual fatty acids in the final sampling. Identical letters, not significantly different. Where no letters are indicated, no significant differences were observed (only 18:2n 6, 20:4n 6, 20:5n 3, 22:5n 3 and 22:6n 3 are tested for significant differences).

9 Fatty acid effects on salmon brain and eye 183 Figure 3 Score plot (PC1 vs. PC2) of the fatty acid composition data of brain (3A) and eye (3B) from the five samplings and the diet fatty acid data (Fe1 to Fe4) from the four dietary groups, and of liver (3C) and fillet (3D) from the four dietary groups from the final sampling, and of the lipid classes from brain (3E) and eye (3F) from group 1 and group 4 from the final sampling. Scales differ.

10 184 T. Brodtkorb et al. diets. Although diets 1 3 contained more 18:2n 6 than 22:6n 3 and all four diets contained more 20:5n 3 than 22:6n 3, the ratio of 18:2n 6 / DHA and DHA / EPA were reversed in the liver of all the four groups relative to the diets. Lie et al. (1992) reported that 22:6n 3 was preferred to 20:5n 3 in the TAG liver stores in cod, which might also be the case for Atlantic salmon. Moreover, the reversed 22:6n 3 / 20:5n 3 ratio can be due to the metabolism of 20:5n 3 in liver mentioned earlier. Utilization of 22:6n 3 is also reported for marine fish larvae (Rønnestad et al. 1995). The ratio of 20:4n 6 / 18:2n 6 was fivefold higher in the liver than in the diets. This might be due to a desaturation and elongation of 18:2(n 6) to 20:4(n 6) in the Atlantic salmon liver. The dietary 20:4(n 6) content increased slightly from diet 1 to diet 4, suggesting that the increased 20:4(n 6) / 18:2(n 6) ratio is most likely due to an active oxidation of 18:2(n 6), thus increasing the 20:4(n 6) / 18:2(n 6) ratio. Brain The score plot (PC1 vs. PC2) of the fatty acid data of brain from the five samplings together with the dietary fatty acids showed no separation of the brain samples (Fig. 3A). The brain samples were gathered in one group correlated negatively to the four diets. The shift of the brain samples along PC1 relative to the diets probably relates to the high 22:6n 3 content in brain compared with the respective diets. The score plot also illustrates the very small effect of the dietary fatty acid composition on the fatty acid composition of the brain lipids. There were, however, effects of the increasing dietary levels of 20:5n 3 on the level of these fatty acids in the brain from the final sampling (Table 3). The relative increase in 20:5n 3 content in brain lipids was 24% compared with 2% relative increase in 22:6n 3 content from group 1 to group 4 fed increasing levels of both fatty acids. This indicates the different degrees of incorporation of 20:5n 3 and 22:6n 3 into the brain lipids when the dietary input of these fatty acids are increased. In addition, the 22:5n 3 level in brain lipids was significantly different between the groups (Table (3). Tocher et al. (1992) reported that cultured brain cells from turbot showed no preferential uptake of 22:6n 3 compared with 18:3n 3. Mourente & Tocher (1992), however, found that DHA was incorporated in brain in preference to other fatty acids by turbot. In the present study, 22:6n 3 was the most abundant fatty acid in the brain of the fish in the four dietary groups, although DHA was not the dominating fatty acid in any of the four diets, and the content ranged from 6.3% to 17.9% of the total dietary fatty acid content. The fact that there was no difference in DHA content in brain lipids in the four dietary groups, together with the great abundance of DHA in brain, points to a selective incorporation of DHA into the developing salmon brain. This is in accordance with the results of Wang et al. (1992), who reported a mechanism in brain for selective uptake of 22:6n 3 into brain PL in mammals. In diets 1 to 3, 18:2n 6 was the dominating fatty acid, whereas in diet 4, 20:5n 3 dominated. The 18:2n 6 content in brain, however, was low and similar in all four groups, contributing only 2 5% of the total fatty acid content. Thus, there seems to be discrimination against incorporating this fatty acid into the salmon brain lipids. The four diets contained considerably more 20:5n 3 than 22:6n 3. This ratio was reversed in the salmon brain, which had a 22:6n 3 / 20:5n 3 ratio of 4.3, corresponding to values found in brain of rainbow trout (Tocher & Harvie 1988) and turbot (Mourente & Tocher 1992). Cod brain, however, had a 22:6n 3 / 20:5n 3 ratio of 3 (Tocher & Harvie 1988). These results demonstrate that 22:6n 3 is the preferred n 3 PUFA in fish brain lipids, before 20:5n 3. The fatty acid composition also illustrate that the brain handles n 3 fatty acids with different chain length and degree of desaturation differently. Whereas the incorporation of 22:6n 3 in brain seems to be strictly regulated and not affected by the dietary level, the incorporation of 20:5n 3 is evidently less controlled. The effects of the increasing dietary 20:5n 3 and 22:6n 3 on the level of these fatty acids in brain are illustrated in Fig. 1. Although the amount of 22:5n 3 was similar in the four diets (Table 2), the relative level of this fatty acid in brain differed between the groups at the end of the experiment (Table 3). Chain elongation of 20:5n 3 is suggested as the reason for the increased 22:5n 3 levels in brain. PE was the dominating lipid class in the salmon brain in the four groups, followed by PC, cholesterol and TAG. Similar results were found in cod (Tocher & Harvie 1988) and turbot brain (Mourente & Tocher 1992). The brain lipid class composition of the four dietary groups are also in accordance with mammalian brain lipid class composition (review, Sastry 1985). The fatty acid compositions of the lipid classes in brain were similar, except for 20:5n 3, in group 1 and group 4, which were the diets with the lowest and highest DHA and EPA levels, respectively. This again illustrates the selectivity of the brain; although 20:5n 3 was significantly higher in group 4 compared with group 1. The only fatty acid in the lipid classes to show a significant response to dietary fatty acid levels was 20:5n 3, and the effect was found in all lipid classes, except PI and PS. The score plot (PC1 vs. PC2) of the fatty acid composition data of the lipid classes divided the material into five classes, one for each lipid class (Fig. 3E). This illustrates the similarity in fatty acid composition within each lipid class, except for 20:5n 3, independent of diet, and the difference in fatty acid composition between the different lipid classes. PI differs from the other lipid classes in having high levels of 20:4n 6, which is universal for

11 Fatty acid effects on salmon brain and eye 185 PI. The content of 20:5n 3 exceeds the 22:6n 3 content in brain PI. Similar results were reported for several tissues in cod (Lie & Lambertsen 1991) and halibut, Hippoglossus hippoglossus (L.) (Lie et al. 1993). Bell & Tocher (1989) reported that rainbow trout brain PI is dominated by 18:0 20:5n 3 species, and not 18:0 20:4n 6, which usually are the dominating fatty acid species in PI. The same is reported for cod brain PI (Bell & Dick 1990). This may be of considerable relevance to eicosanoid metabolism in the brain. It has been reported that the DHA supply to the nervous tissue in mammals is ensured through lipoproteins from the liver, comprising an efficient route to meet the demands for the fatty acid by those tissues (Martin et al. 1994). The selectivity of the brain towards DHA demonstrated in the present study may be due to lipoprotein lipase. This endothelial enzyme responsible for hydrolysing TAG in chylomicrons and VLDL is found in particulary high levels in rat brain during postnatal development (Chajek et al. 1977). However, Anderson et al. (1994) found that it is unlikely that lipoprotein lipase alone is responsible for the selective uptake of 22:6n 3. A family of soluble 21- to 23-kd lipid-binding proteins in brain and other tissues was identified (Schoentgen et al. 1993). It is therefore suggested that there exists a selective DHA-receptor in brain, which may bind and/or transport DHA over the blood brain barrier. There are indications of the existence of such a selective DHA-binding protein in retina of rat (Lee et al. 1995), which supports the theory of a similar DHA-binding protein in brain. These binding proteins are, however, found in vitro in the soluble, cytosolic fraction in the rat retina (Lee et al. 1995). Because of this, the authors assumed that the DHA-binding protein could serve as a carrier in the transport or metabolic processing of DHA rather than as a terminal acceptor of the fatty acid. Considerable amounts of DHA were found in cholesterol ester (CE), probably in HDL, in mice serum (Martin et al. 1994). This suggests a role of CE in DHA transport. Furthermore, the apoprotein, apod, which is present in HDL in complex with lecithincholesterol acyltransferase (LCAT), is also found in brain tissue in the rhesus monkey, mainly in perivascular areas. ApoD may be involved in the influx and eflux of cholesterol and of CE containing high levels of DHA, both of which are needed during brain development, over the blood-brain barrier (Smith et al. 1990). Eye The score plot (PC1 vs. PC2) of the fatty acid composition data from the eye samples and from the four diets gave no clear separation of the data (Fig. 3B). The four diets are, however, separated from the eye samples. The four experimental diets were assembled on a line with diet 1 positively correlated to 18:2n 6, which had a high negative loading along PC1, and diets 2, 3 and 4 being gradually less positively correlated to 18:2n 6 (indicated by an arrow in Fig. 3B). The 22:6n 3, which had a positive loading along PC1, also affect the arrangement of the four diets, with diet 4 most positively correlated to 22:6n 3. There is, however, a dispersion of the eye samples in the score plot which demonstrates the response of eye lipids to dietary intake. The significantly different levels of 18:2n 6, 20:5n 3 and 22:5n 3 between the groups at the end of the experiment (Table 3) demonstrates the effect of the diets on the 20:5n 3 and 18:2n 6 levels in eye. The relative content of 22:6n 3 in eye was, as in brain, not significantly affected by the different dietary levels. The relative increase in 22:6n 3 content of 29% from groups 1 to 4, however, shows that there was a minor effect of the dietary DHA on the relative eye DHA content. The 20:5n 3 was clearly affected by the increasing dietary levels, with a relative increase of 54% from group 1 to group 4, which suggests that the different handling of 20:5n 3 and 22:6n 3 is similar in brain and eye. The diet-dependent levels of 22:5n 3 found in eye in the final sampling is probably due to chain elongation of 20:5n 3, as in brain. Also, the 22:6n 3 / 20:5n 3 ratio was reversed in the eye compared with the respective diets, similarly as in brain. The relative content of 22:6n 3 decreased during the feeding period, due to a concomitant significant increase in 18:2n 6. The eye is therefore, as mentioned earlier, not as selective as brain concerning the incorporation of 18:2n 6. TAG was the predominating lipid class in the salmon eye, comprising as much as 70% of the total lipid classes. This amounted to mg TAG g 1 wet weight eye, whilst the other lipid classes did not exceed 5 mg g 1. Similar results were reported by Tocher & Harvie (1988), who found significantly higher amounts of TAG in the retina of rainbow trout compared with cod. The PE and PC levels found in the salmon eye were slightly lower than amounts found in retinas of adult rainbow trout and cod (Tocher & Harvie 1988). This may be due to species difference or, more likely, to the different analytical methods used for determining the lipid class contents. DHA was the most abundant fatty acid in all the phosphoacylglycerols in the eye (Table 6). There was significantly more 20:5n 3 in eyes from group 4 compared with those from group 1 in all five lipid classes. Except for neutral lipids, which reflect the fatty acid compositions of the two diets, the fatty acid composition of the phospholipids showed only minor response to the dietary fatty acid composition, with 20:5n 3 as the only exception. Lie et al. (1992) reported that an incorporation of 18:2n 6 into PC of fillet, heart and gill in cod was not fully balanced by a decrease of 20:5n 3 and 22:6n 3. They suggested that 18:2n 6 might be incorporated into PC as an alternative to monoenes or

12 186 T. Brodtkorb et al. saturated fatty acids in cod. The bulk of 18:2n 6 in salmon eye was incorporated into the neutral lipids, not in PC or any of the other phospholipids, which may indicate that 18:2n 6 was not incorporated as a replacement for monoene fatty acids in the phospholipids in salmon eye. The score plot (PC1 vs. PC2) of the fatty acid composition data of the lipid classes divided the data into five classes, one for each lipid class (Fig. 3F). The fatty acid composition of the different phospholipids in the eye seems to be independent of the dietary fatty acid composition, except for 20:5n 3. PI in the eye was characterized by a high 20:4n 6 content and a relatively low n 3 PUFA content compared with the other phospholipids. It is noteworthy that PI in the eye contains more 20:4n 6 than 20:5n 3, as in trout and cod retina, which is opposite to the situation in brain PI in the developing salmon. Di-22:6n 3 phospholipids are reported to be abundant in fish retina (Bell & Dick 1991). PE in the salmon eye contained almost 60% 22:6n 3, which indicates the existence of di-22:6n 3 molecular species. DHA was the most abundant fatty acid in the salmon eye in all the dietary groups in spite of the different fatty acid composition of the diets. This suggests a selective uptake of 22:6n 3 into the developing salmon retina, which has also been demonstrated for rat retina (Wang et al. 1992). A study on frogs, Rana pipiens, revealed that the retinal pigment epithelium (RPE) has the ability to selectively and efficiently take up DHA from plasma (Bazan et al. 1994). The mechanism in the selective uptake of DHA by RPE is not established. However, a recent publication (Lee et al. 1995) indicated the existence of fatty acid binding proteins in rat retina in vitro. Although the proteins are not properly identified, the results reported by Lee et al. (1995) strongly suggest the existence of a unique DHA-binding protein (43 kd) in the retina. The RPE is located between the rod outer segments (ROS) and the choriocapillars, the blood supply for the photoreceptor cells, where receptors for low-density lipoprotein (LDL) have been found in bovines (Hayes et al. 1989). It is possible that it is here, in the transport of DHA through the RPE, that the specific DHAbinding proteins function. When fed 22:6n 3-deficient diets, the eyes of mammals (Crawford 1990) and fish (Sargent et al. 1993) and the brains of rodents (Wainwright 1992) and of fish (Mourente & Tocher 1992) showed reduced 22:6n 3 levels and concomitant deficiency symptoms. Evidently, to produce major changes in fatty acid composition of brain and eye, a DHA deficiency must be present. The growth and mortality data in this experiment show that the dietary levels of DHA and EPA were sufficient for normal growth and development, and furthermore these levels of 22:6n 3 and 20:5n 3 gave only minor changes in fatty acid composition in brain and eye. The fatty acid composition of liver and fillet, however, was clearly affected by the dietary fatty acid composition. This demonstrates the possibility of designing salmon fillet with a desired fatty acid composition. Acknowledgements We would like to thank Nutreco Aquaculture Research Centre, Stavanger, for financial support. The skilled technical assistance of Kjersi Ask, Annbjørg Eliassen and Kari-Elin Langeland is gratefully acknowledged. Furthermore, Kjellaug R. Ingebrigtsen and Aslaug Berge are acknowledged for invaluable help at Nutreco Aquaculture Research Station. References Albrektsen, S., Hagure, T.A. & Lie, Ø. (1994) The effect of dietary vitamin B6 on tissue fat and lipid composition in livers and gills in Atlantic salmon (Salmo salar). Comp. Biochem. Physiol., 109A, Anderson, G. & Connor, W.E. (1988) Uptake of fatty acids by developing rat brain. Lipids, 23, Anderson, G.J., Tso, P.S. & Connor, W.E. (1994) Incorporation of chylomicron fatty acids into the developing rat brain. J. Clin. Invest., 93, Bazan, N.G., Rodriguez, B.E. & Gordon, W.C. (1994) Docosahexaenoic acid supply to the retina and its conservation in photoreceptor cells by active retinal pigment epithelium-mediated recycling. World Rev. Nutr. Diet, 75, Bell, M.V. & Dick, J.R. (1991) Molecular species composition of the major diacyl glycero-phospholipids from muscle, liver, retina and brain of cod (Gadus morhua). Lipids, 26, Bell, M.V. & Dick, J.R. (1990) Molecular species composition of phosphatidylinositol from the brain, retina, liver and muscle of cod (Gadus morhua). Lipids, 25, Bell, M.V. & Tocher, D.R. (1989) Molecular species composition of the major phospholipids in brain and retina from rainbow trout (Salmo gairdneri). Biochem. J., 264, Bell, M.V., Batty, R.S., Dick, J.R., Fretwell, K., Navarro, J.C. & Sargent, J.R. (1995) Dietary deficiency of docosahexaenoic acid impairs vision at low light intensities in juvenile herring (Clupea harengus L.). Lipids, 30, Chajek, T., Stein, O. & Stein, Y. (1977) Pre- and post-natal development of lipoprotein lipase and hepatic triglyceride hydrolase activity in rat tissues. Atherosclerosis, 26, Connor, W.E., Neuringer, M. & Reisbick, S. (1992) Essential fatty acids: the importance of n 3 fatty acids in the retina and brain. Nutr. Rev., 50(4), Crawford, M.A. (1990) The early development and evolution of the human brain. Upsala J. Med. Sci., Suppl., 48, Hayes, K.C., Lindsey, S., Stephan, Z.F. & Brecker, D. (1989) Retinal pigment epithelium possesses both LDL and scavenger receptor activity. Invest. Ophthalmol. Vis. Sci., 30, Innis, S.M. (1991) Essential fatty acids in growth and development. Prog. Lipid Res., 30, Lee, J., Jiao, X., Gentleman, S., Wetzel, M.G., O Brien, P. & Chader, G.J. (1995) Soluble-binding proteins for docosahexaenoic acid are present in neural retina. Invest. Ophthalmol. Vis. Sci., 36, Lie, Ø. & Lambertsen, G. (1991) Fatty acid composition of glycerophospholipids in seven tissues of cod (Gadus morhua), determined by combined high-performance liquid chromatography and gas chromatography. J. Chromatography, 565,